ehlo,
There were two different crashed in BZ1079237 [1] The first one was in sssd_be (it was fixed commit e18d5c703c529f99bd375da887d9) and the second one in sss autofs plugin. (client code)
Attached patches should fix the sond crash.
How to reproduce? a) -- revert commit e18d5c703c529f99bd375da887d9 -- have more then one attribute "automountKey" in LDAP entry -- call automount -m b) #can be reproduced with current master -- use small value of option timeout in autofs section (can be default) -- attach gdb to the autof responder -- add breakpoint to the function b setautomntent_send and continue -- call automount -m -- gdb in autofs responder will interrupt execution (do nothing) -- automount will crash after few seconds (60-120) LS
On Tue, May 27, 2014 at 03:52:23PM +0200, Lukas Slebodnik wrote:
ehlo,
There were two different crashed in BZ1079237 [1] The first one was in sssd_be (it was fixed commit e18d5c703c529f99bd375da887d9) and the second one in sss autofs plugin. (client code)
Attached patches should fix the sond crash.
How to reproduce? a) -- revert commit e18d5c703c529f99bd375da887d9 -- have more then one attribute "automountKey" in LDAP entry -- call automount -m b) #can be reproduced with current master -- use small value of option timeout in autofs section (can be default) -- attach gdb to the autof responder -- add breakpoint to the function b setautomntent_send and continue -- call automount -m -- gdb in autofs responder will interrupt execution (do nothing) -- automount will crash after few seconds (60-120)
Reproduced and verified the fix using this ^^ method. The code looks good to me as well and follows the pattern in other responders now
ACK to both.
LS
On Tue, May 27, 2014 at 05:34:22PM +0200, Jakub Hrozek wrote:
On Tue, May 27, 2014 at 03:52:23PM +0200, Lukas Slebodnik wrote:
ehlo,
There were two different crashed in BZ1079237 [1] The first one was in sssd_be (it was fixed commit e18d5c703c529f99bd375da887d9) and the second one in sss autofs plugin. (client code)
Attached patches should fix the sond crash.
How to reproduce? a) -- revert commit e18d5c703c529f99bd375da887d9 -- have more then one attribute "automountKey" in LDAP entry -- call automount -m b) #can be reproduced with current master -- use small value of option timeout in autofs section (can be default) -- attach gdb to the autof responder -- add breakpoint to the function b setautomntent_send and continue -- call automount -m -- gdb in autofs responder will interrupt execution (do nothing) -- automount will crash after few seconds (60-120)
Reproduced and verified the fix using this ^^ method. The code looks good to me as well and follows the pattern in other responders now
ACK to both.
Pushed to master: 01f0d708fddfd50bdbb36a63737dcdc2c2fbd28d 71bcdde36bcfd8cb797777062208a85bf6337dec and sssd-1-11: 03d7e3207a8758130d7e2431cd886bb3edfc6190 5fc854880db68d9fb887b2d2767906dbf3a1ebe6
sssd-devel@lists.fedorahosted.org
-
Jakub Hrozek -
Lukas Slebodnik